Development and application of accident prediction model for intersections at Dhaka city

Abstract

Road accidents nowadays have become the major cause of deaths for its severity causing injury, fatal causalities, property damage, death of millions of people. Commencing with the data that every year more than 1.3 million people die in road accidents around the world (Building Leaders in Urban Transport Planning, LUTP, 2012). So, this indicates that the major mortality rate is procured by the road accidents no question asked. Not only in USA or any other developed countries of the world, but also in developing country like Bangladesh, with regard to safety, the pedestrians are of major concern as they represent up to 72 percent of road traffic fatalities in Dhaka Metropolitan City (Rahman et al., 2006). Talking about Dhaka City, data show that the intersection accidents represent around 40 percent of total accidents occurring in Metropolitan City of Dhaka (Rahman, 2012). So, it can be stated that pedestrian safety is not only a problem in global range, but also acute in developing country like Bangladesh. However, to guide in making improved policy statement or legal instructions, our study tends to reflect the prediction of accident occurring with the presence of different set of variables taken at selected accident prone intersections in Dhaka City. The study is especially designed to predict accident probability only for pedestrian-vehicle crashes at intersections as accurate as possible. Variables set were taken in geometric and regulatory types. So, the principal objective of this study is to develop a good statistical model to set up a suitable relationship between accident occurrence and intersection geometric and regulatory control characteristics. Another objective is to compare some of the results of our study with the parameters of pedestrian-vehicle crashes of developed countries to get an evaluative situation easily understandable for all. While the methodology may apply to any traffic accident at intersections on any type of roadways. The statistical models developed in this study are focused in predicting traffic accidents at four-legged signalized and Tee or three legged intersections. For data analyzing and modeling of the pedestrian-vehicle crashes at intersection, at first annual crash data were collected from 1998-2009 from ARI(Accident Research Institute), BUET. There is a total of 54 intersection‟s data were available and we worked on 45 selected intersections. Hand drawn pictures were made for all of this intersections for getting geometric data precisely. After getting all the data analyzed by the statistical software, a projected equation is possible to make with the help of the critical and significant factors to get accident prediction for accident forecasting at any particular intersection. Which factor will have increased impact on traffic and pedestrian iv crashes and which will reduce impact to accident occurrence, there will be a clear picture of it. Now that, if anyone ask about what is the possibility of accident occurrence at Cross section of Kakoli (Mymensingh Road+ Kamal Atarturk Avenue) with the information saying road width 30 ft, one-way, signalized intersection, two bus stops at upstream and downstream of the Dhaka-Mymensingh. Road, U-turn prohibited, median & roadside barrier present, VMS said:vehicle speed <40mph(just any sign, not that speed is considered only. Actually the VMS presence in any form is taken into consideration). So, with the above information, it is possible to predict the accident occurrence scenario in regular, slow, thorough manner and what can be done to reduce the crash possibility at this intersection by comparing the info with our significant accident factors characteristics from the model result. Here, the important thing is that the factors which are not present in our significant list can be omitted saying that it has no such effect on accident occurring. However, by taking all these variables into consideration it will be possible to draw the outline that which factors are influential for which intersections and which aren‟t, out of the 45 intersections we covered in Dhaka City based on the previous accident prone intersection‟s map, data and studies.